LES channelFoam solver underpredicts friction factor
 

I am running a (periodic) pipe flow calculation using channelFoam and am unable to obtain an appropriate value for the friction factor. The case is wall-resolved, with wall-normal, span-wise and stream-wise cell dimensions (given in wall units) of 0.5, 30 and 50, respectively, with a growth rate of 1.2 within the prism layer. The Reynolds number in my test case is 50,000, and the length-to-diameter ratio, L/D, is equal to 6.

The Blausius relationship for the friction factor gives a value of f = 0.0022, but my calculation is giving a value of approximately 0.0016. I am using Alberto Passalacqua's implementation of the dynamic Smagorinsky model. Is there an inherent issue that I am overlooking in applying the dynamic model for a wall-resolved case such as this? I am using an unstructured mesh with wedges in the core and hexahedrons in the prism layers. I have tried several different LES models and filters and meshes, but nothing seems to improve the result.

I am aware of the fact that the use of an unstructured, non-uniform mesh can be an issue given the non-commutative nature of the filters, but am not sure whether or not the underprediction of the wall shear stress is an expected consequence. Does anyone have any thoughts on this? Any suggestions?